AAV vectors may have utility for gene therapy but heretofore a significant obstacle has been the inability to generate sufficient quantities of such recombinant vectors in amounts that would be clinically useful for human gene therapy application. This is a particular problem for in vivo applications such as direct delivery to the lung.
Adeno-associated virus (AAV) vectors are among a small number of recombinant virus vector systems which have been shown to have utility as in vivo gene transfer agents (reviewed in Carter, 1992, Current Opinion in Biotechnology, 3:533-539; Muzcyzka, 1992, Curr. Top. Microbiol. Immunol. 158:97-129) and thus are potentially of great importance for human gene therapy. AAV vectors are capable of high-frequency stable DNA integration and expression in a variety of cells including cystic fibrosis (CF) bronchial and nasal epithelial cells (Flotte et al., 1992a, Am. J. Respir. Cell Mol. Biol. 7:349-356; Egan et al., 1992, Nature, 358:581-584; Flotte et al., 1993a, J. Biol. Chem. 268:3781-3790; Flotte et al., 1993b, Proc. Natl. Acad. Sci. USA, in press), human bone marrow-derived erythroleukemia cells (Walsh et al., 1992, Proc. Natl. Acad. Sci. USA, 89:7257-7261), and several others. AAV may not require active cell division for stable expression which would be a clear advantage over retroviruses, especially in tissue such as the human airway epithelium where most cells are terminally differentiated and non-dividing.
AAV is a defective parvovirus that grows only in cells in which certain functions are provided by a co-infecting helper virus (see FIG. 1). General reviews of AAV may be found in Carter, 1989, Handbook of Parvoviruses, Vol. I, pp. 169-228, Carter, 1989, Handbook of Parvoviruses, Vol. I, pp. 169-228, Berns, 1990, Virology, pp. 1743-1764, Raven Press, (New York). Examples of co-infecting viruses that provide helper functions for AAV growth and replication are adenoviruses, herpesviruses and in some cases poxviruses such as vaccinia. The nature of the helper function is not known but appears to be some indirect effect of the helper virus which renders the cell permissive for AAV replication. This concept is supported by the observation that in certain cases AAV replication may occur at a low level of efficiency in the absence of helper virus co-infection if the cells are treated with agents that are either genotoxic or that disrupt the cell cycle.
Although AAV may replicate to a limited extent in the absence of helper virus in certain unusual conditions, as noted above, the more general result is that infection of cells with AAV in the absence of helper functions results in integration of AAV into the host cell genome. The integrated AAV genome may be rescued and replicated to yield a burst of infectious progeny AAV particles if cells containing an integrated AAV provirus are superinfected with a helper virus such as adenovirus. Because the integration of AAV appears to be an efficient event, this suggested that AAV would be a useful vector for introducing genes into cells for stable expression for uses such as human gene therapy. More recent results (Kotin & Berns, 1989, Virology 170:460-467; Kotin et al., 1990, Proc. Natl. Acad. Sci. USA, 87:2211-2215; Samulski et al., 1991, EMBO J. 10:3941-3950) have suggested that AAV may exhibit some preference for integration at a site on human chromosome 19 but the generality and mechanism of this phenomenon has not been elucidated fully.
AAV has a very broad host range with neither any obvious species or tissue specificity and will replicate in virtually any cell line of human, simian or rodent origin provided an appropriate helper is present. AAV is ubiquitous and has been isolated from a wide variety of animal species including most mammalian and several avian species.
AAV has not been associated with the cause of any disease. AAV is not a transforming or oncogenic virus. AAV integration into chromosomes of human cell lines does not cause any significant alteration in the growth properties or morphological characteristics of the cells. These properties of AAV also recommend it as a potentially useful human gene therapy vector because most of the other viral systems proposed for this application such as retroviruses, adenoviruses, herpesviruses, or poxviruses are disease-causing viruses.
AAV particles are comprised of a protein capsid having three capsid proteins, VP1, VP2, and VP3, and enclosing a DNA genome. The AAV DNA genome is a linear single-stranded DNA molecule having a molecular weight of about 1.5.times.10.sup.6 daltons or approximately 4680 nucleotides long. Strands of either complementary sense, "plus" or "minus" strands, are packaged into individual particles but each particle has only one DNA molecule. Equal numbers of AAV particles contain either a plus or minus strand. Either strand is equally infectious and replication occurs by conversion of the parental infecting single strand to a duplex form and subsequent amplification of a large pool of duplex molecules from which progeny single strands are displaced and packaged into capsids. Duplex or single-strand copies of AAV genomes inserted into bacterial plasmids or phagemids are infectious when transfected into adenovirus-infected cells, and this has allowed the study of AAV genetics and the development of AAV vectors. The replication cycle of AAV is diagrammed in FIG. 1.
The AAV2 genome has one copy of the 145-nucleotide-long ITR (inverted terminal repeat) of each end and a unique sequence region of about 4470 nucleotides long (Srivastava et al., 1983, J. Virol, 45:555-564) that contains two main open reading frames for the rep and cap genes (Hermonat et al., J. Virol. 51:329-339; Tratschin et al., 1984a, J. Virol., 51:611-619). The unique region contains three transcription promoters p.sub.5, p.sub.19, and p.sub.40 (Laughlin et al., 1979, Proc. Natl. Acad. Sci. USA, 76:5567-5571) that are used to express the rep and cap genes. The ITR sequences are required in cis and are sufficient to provide a functional origin of replication (ori) and also are sufficient to provide signals required for integration into the cell genome as well as for efficient excision and rescue from host cell chromosomes or from recombinant plasmids. In addition it has been shown that the ITR can function directly as a transcription promoter in an AAV vector (Flotte et al., 1993, vide supra).
The rep and cap genes are required in trans to provide functions for replication and encapsidation of vital genome respectively. The rep gene is expressed from two promoters, p.sub.5 and p.sub.19. Transcription from p.sub.5 yields an unspliced 4.2 kb mRNA which encodes a protein, Rep78, and a spliced 3.9 kb mRNA which encodes a protein, Rep68. Transcription from p.sub.19 yields an unspliced mRNA which encodes Rep52 and a spliced 3.3 kb mRNA which encodes Rep40. Thus, the four Rep proteins all comprise a common internal region sequence but differ with respect to their amino and carboxyl terminal regions. Only Rep78 and Rep68 are required for AAV duplex DNA replication, but Rep52 and Rep40 appear to be needed for progeny, single-strand DNA accumulation. Mutations in Rep78 and Rep68 are phenotypically Rep.sup.- whereas mutations affecting only Rep52 and Rep40 are Rep.sup.+ but Ssd.sup.-. Rep68 and Rep78 bind specifically to the hairpin conformation of the AAV ITR and possess several enzyme activities required for resolving replication at the AAV termini. Rep52 and Rep40 have none of these properties.
The Rep proteins, primarily Rep78 and Rep68 exhibit several pleiotropic regulatory activities including positive and negative regulation of AAV genes and expression from some heterologous promoters, as well as inhibitory effects on cell growth (Tratschin et al., 1986, Mol. Cell. Biol. 6:2884-2894; Labow et al., 1987, Mol. Cell. Biol., 7:1320-13251Khleif et al., Virology, 181:738-741). The AAV p.sub.5 promoter is negatively autoregulated by Rep78 or Rep68 (Tratschin et al., 1986, Mol. Cell. Biol. 6:2884-2894). Because of the inhibitory effects of expression of rep on cell growth, constitutive expression of rep in cell lines has not been readily achieved. For example, Mendelson et al. (1988, Virology, 166:154-165) reported a very low level expression of some Rep proteins in certain cell lines after stable integration of AAV genomes.
The proteins VP1, VP2, and VP3 all share a common overlapping sequence but differ in that VP1 and VP2 contain additional amino terminal sequence. All three are coded from the same cap gene reading frame expressed from a spliced 2.3 kb mRNA transcribed from the p.sub.40 promoter. VP2 and VP3 are generated from the same mRNA by use of alternate initiation codons. VP1 is coded from a minor mRNA using 3' donor site that is 30 nucleotides upstream from the 3' donor used for the major mRNA that encodes VP2 and VP3. VP1, VP2, and VP3 are all required for capsid production. Mutations which eliminate all three proteins (Cap.sup.-) prevent accumulation of single-strand progeny AAV DNA whereas mutations in the VP1 amino-terminus (Lip.sup.-, Inf.sup.-) permit single-strand production but prevent assembly of stable infectious particles.
The genetic analysis of AAV that was described above was based upon mutational analysis of AAV genomes that were molecularly cloned into bacterial plasmids. In early work, molecular clones of infectious genomes of AAV were constructed by insertion of double-strand molecules of AAV into plasmids by procedures such as GC tailing (Samulski et al., 1982, Proc. Natl. Acad. Sci. USA, 79:2077-2081), addition of synthetic linkers containing restriction endonuclease (Laughlin et al., 1983, Gene, 23:65-73) or by direct, blunt-end ligation (Senapathy & Carter, 1984, J. Biol. Chem., 259:4661-4666). It was then shown that transfection of such AAV recombinant plasmids into mammalian cells that were also infected with an appropriate helper virus, such as adenovirus, resulted in rescue and excision of the AAV genome free of any plasmid sequence and replication of the rescued genome and generation of a yield of progeny infectious AAV particles (see FIG. 1). This provided the basis for performing genetic analysis of AAV as summarized above and permitted construction of AAV transducing vectors.
Based on the genetic analysis, the general principles of AAV vector construction were defined as reviewed recently (Carter, 1992, Current Opinions in Biotechnology, 3:533-539; Muzyczka, 1992, Current Topics in Microbiology and Immunology, 158:97-129). AAV vectors are constructed in AAV recombinant plasmids by substituting portions of the AAV coding sequence with foreign DNA to generate a vector plasmid. In the vector plasmid, the terminal (ITR) portions of the AAV sequence must be retained intact because these regions are required in cis for several functions including excision from the plasmid after transfection, replication of the vector genome and integration and rescue from a host cell genome. The vector can then be packaged into an AAV particle to generate an AAV transducing virus by transfection of the vector plasmid into cells that are infected by an appropriate helper virus such as adenovirus or herpesvirus. In order to achieve replication and encapsidation of the vector genome into AAV particles, the vector plasmid must be complemented for any AAV functions required in trans, namely rep and cap, that were deleted in construction of the vector plasmid.
There are at least two desirable features of any AAV vector that is designed for use in human gene therapy. First, the transducing vector must be generated at sufficiently high titers that it is practicable as a delivery system. This is especially important for gene therapy stratagems aimed at in vivo delivery of the vector. It is likely that for many desirable applications of AAV vectors, such as treatment of cystic fibrosis by direct in vivo delivery to the airway, the required dose of transducing vector may be in excess of 10.sup.10. Secondly, the vector preparations must be free of wild-type AAV virus. The attainment of high titers of AAV vectors has been difficult for several reasons including preferential encapsidation of wild-type AAV genomes if they are present or generated by recombination, and the inability to generate sufficient complementing functions such as rep or cap. Useful cell lines expressing such complementing functions have not been generated, in part, because of several inhibitory functions of the rep gene.
The first AAV vectors that were described contained foreign reporter genes such as neo or cat or dhfr that were expressed from AAV transciption promoters or an SV40 promoter (Tratschin et al., 1984b, Mol. Cell. Biol. 4:2072-2081; Hermonat & Muzyczka, 1984, Proc. Natl. Acad. Sci. USA, 81:6466-6470; Tratschin et al., 1985, Mol. Cell. Biol. 5:3251-3260; McLaughlin et al., 1988, J. Virol., 62:1963-1973; Lebkowski et al., 1988 Mol. Cell. Biol., 7:349-356). These vectors were packaged into AAV-transducing particles by co-transfection into adenovirus-infected cells together with a second packaging plasmid that contained the AAV rep and cap genes expressed from the natural wild-type AAV transciption promoters. In an attempt to prevent packaging of the packaging plasmid into AAV particles several approaches were taken. In some cases, (Hermonat & Muzyczka, 1984; McLaughlin et al., 1988) the packaging plasmid had inserted a large region of bacteriophage lambda DNA within the AAV sequence to generate an oversized genome that could not be packaged. In other cases, (Tratschin et al., 1984b; Tratschin et al., 1985, Lebkowski et al., 1988), the packaging plasmid had deleted the ITR regions of AAV in order that it could not be excised and replicated and thus could not be packaged. All of these approaches failed to prevent generation of particles containing wild-type AAV DNA and also failed to generate effective high titere of AAV transducing particles. Indeed titers of not more than 10.sup.4 ml were cited by Hermonat & Muzyczka, 1984. The production of wild-type AAV particles in these studies was probably due to the presence of overlapping homology between AAV sequences present in the vector and packaging plasmids. It was shown by Senapathy and Carter (1984, J. Biol. Chem. 259:4661-4666) that the degree of recombination in such a system is approximately equivalent to the degree of sequence overlap. It was suggested in a review of the early work (Carter 1989, Handbook of Parvoviruses, Vol. II, pp. 247-284, CRC Press, Boca Raton, Fla.) that titers of 10.sup.6 per ml might be obtained, but this was based on the above-cited studies in which large amounts of wild-type AAV contaminated the vector preparation. Such vector preparations containing wild-type AAV are not useful human gene therapy. Furthermore, these early vectors exhibited low transduction efficiencies and did not transduce more than 1 or 2% of cells in cultures of various human cell lines even though the vectors were supplied at multiplicities of up to 50,000 particles per cell. This may have reflected in part the contamination with wild-type AAV particles and the presence of the AAV rep gene in the vector. Furthermore, Samulski et al. (1989, J. Virol. 63:3822-3828) showed that the presence of wild-type AAV significantly enhanced the yield of packaged vector. Thus, in packaging systems where the production of wild-type AAV is eliminated, the yield of packaged vector may actually be decreased. Nevertheless, for use in any human clinical application it will be essential to eliminate production of wild-type AAV.
Additional studies (McLaughlin et al., 1988; Lebkowski et al., 1988) to generate AAV vectors which did not contain the AAV rep or cap gene still met with generation of wild-type AAV and still produced very low transduction frequencies on human cell lines. Thus, McLaughlin et al., 1988 reported that AAV rep.sup.- cap.sup.- vectors containing the neo gene packaged with the same packaging plasmid used earlier by Hermonat & Muzyczka (1984) still contained wild-type AAV. As a consequence it was only possible to use this virus at a multiplicity of 0.03 particles per cell (i.e., 300 infectious units per 10,000 cell) to avoid double hits with vector and wild-type particles. When the experiment was done in this way, by infecting 32,000 cells with 1000 infectious units, an average of 800 geneticin-resistant colonies was obtained. Although this was interpreted as demonstrating the virus was capable of yielding a transduction frequency of 80%, in fact only 2.5% of the cells were transduced. Thus the effectively useful titer of this vector was limited. Furthermore, this study did not demonstrate that the actual titer of the vector preparation was any higher than those obtained previously by Hermonat & Muzyczka (1984). Similarly, Lebkowski et al., 1988, packaged AAV vectors which did not contain either a rep or cap gene and used an ori.sup.- packaging plasmid pBa1A identical to that used earlier by Tratschin et al., (1984b, 1985) and reported transduction frequencies that were similarly low, in that for several human cell lines not more than 1% of the cells could be transduced to geneticin resistance even with their most concentrated vector stocks. Lebkowski et al., (1988) did not report the actual vector titers in a meaningful way but the biological assays showing not more than 1% transduction frequency when 5.times.10.sup.6 cells were exposed to three ml of vector preparation indicates that the titer was less than 2.times.10.sup.4. Also, the pBa1 packaging plasmid contains overlapping homology with the ITR sequence in the vector and leads to generation by recombination of wild-type AAV.
Laface et al., (1988) used the same vector as that used by Hermonat & Muzyczka (1984) prepared in the same way and obtained a transduction frequency of 1.5% in murine bone marrow cultures again showing very low titer.
Samulski et al., (1987, J. Virol., 61:3096-3101) constructed a plasmid called pSub201 which was an intact AAV genome in a bacterial plasmid but which had a deletion of 13 nucleotides at the extremity of each ITR and thus was rescued and replicated less efficiently than other AAV plasmids that contained the entire AAV genome. Samulski et al. (1989, J. Virol., 63:3822-3828) constructed AAV vectors based on pSub201 but deleted for rep and cap and containing either a hyg or neo gene expressed from an SV40 early gene promoter. They packaged these vectors by co-transfection with a packaging plasmid called pAAV/Ad which consisted of the entire AAV nucleotide sequence from nucleotide 190 to 4490 enclosed at either end with one copy of the adenovirus ITR. In this packaging plasmid the AAV rep and cap genes were expressed from the natural AAV promoters p.sub.5, p.sub.19 and p.sub.40. The function of the adenovirus ITR in pAAV/Ad was thought to be to enhance the expression level of AAV capsid proteins. However, rep is expressed from its homologous promoter and is negatively regulated and thus its expression is limited. Using their encapsidation system Samulski et al., 1989, generated AAV vector stocks that were substantially free of wild-type AAV but had transducing titers of only 3.times.10.sup.4 hygromycin-resistant units per ml of supernatant. When a wild-type AAV genome was used in the packaging plasmid the titer of the AAV vector prep was increased to 5.times.10.sup.4. The low titer produced in this system thus appears to have been due in part to the defect in the ITR sequences of the basic pSub201 plasmid used for vector construction and in part due to limiting expression of AAV genes from pAAV/Ad. In an attempt to increase the titer of the AAVneo vector preparation, Samulski et al., 1989, generated vector stocks by transfecting, in bulk, thirty 10-cm dishes of 293 cells and concentrating the vector stock by banding in CsCl. This produced an AAVneo vector stock containing a total of 10.sup.8 particles as measured by a DNA dot-blot hybridization assay. When this vector stock was used at multiplicities of up to 1,000 particles per cell, a transduction frequency of 70% was obtained. This suggests that the particle-to-transducing ratio is about 500 to 1,000 particles since at the ratio of one transducing unit per cell the expected proportion of cells that should be transduced is 63% according to the Poisson distribution.
Although the system of Samulski et al., 1989, using the vector plasmid pSub201 and the packaging plasmid pAAV/Ad did not have overlapping AAV sequence homology between the two plasmids, there is overlapping homology at the XbaI sites and recombination of these sites leads to generation of complete wild-type AAV. That is, although overlapping homology of AAV sequence is not present, the complete AAV sequence is contained within the two plasmids, and thus recombination can generate wild-type AAV, which is undesirable. That this class of recombination occurs in AAV plasmids was shown by Senapathy & Carter (1984, J. Biol. Chem. 259:4661-4666). Therefore, because of the problems of low titer and ability to generate wild-type recombinants, the system described by Samulski et al., 1989, does not have utility for human gene therapy.
Several other reports have described AAV vectors. Srivastava et al., (1989, Proc. Natl. Acad. Sci. USA, 86:8078-8082) described an AAV vector based on the pSub201 plasmid of Samulski et al., (1987), in which the coding sequences of AAV were replaced with the coding sequences of another parvovirus, B19. This vector was packaged into AAV particles using the pAAV/Ad packaging plasmid and generated a functional vector, but titers were not reported. This system was based on pSub281 and thus suffers from the defect described above for this plasmid. Second, the vector and the packaging plasmid both contained overlapping AAV sequences (the ITR regions) and thus recombination to give contaminating wild-type virus is highly likely.
Chatterjee et al. (1991, Vaccines 91, Cold Spring Harbor Laboratory Press, pp. 85-89), Wong et al. (1991 Vaccines 91, Cold Spring Harbor Laboratory Press, pp. 183-189), and Chatterjee et al. (1992, Science, 258:1485-1488) describe AAV vectors designed to express antisense RNA directed against infectious viruses such as HIV or Herpes simplex virus. However, these authors did not report any titers of their AAV vector stocks. Furthermore, they packaged their vectors using and Ori.sup.- packaging plasmid analogous to that used by Tratschin et al. (1984b, 1985) containing the Ba1A fragment of the AAV genome and therefore their packaging plasmid contained AAV vector sequences that have homology with AAV sequences that were present in their vector constructs. This will also lead to generation of wild-type AAV. Thus, Chatterjee et al., and Wong et al., used a packaging system known to give only low titer and which can lead to generation of wild-type AAV genomes because of the overlapping homology in the vector and packaging sequences.
Other reports have described the use of AAV vectors to express genes in human lymphocytes (Muro-Cacho et al., 1992, J. Immunotherapy, 11:231-237) or a human erythroid leukemia cell line (Walsh et al., 1992, Proc. Natl. Acad. Sci. USA, 89:7257-7261) with vectors based on the pSub201 vector plasmid and pAAV/Ad packaging plasmid. Again, titers of vector stocks were not reported and were apparently low because a selective marker gene was used to identify those cells that had been successfully transduced with the vector.
Transduction of human airway epithelial cells, grown in vitro from a cystic fibrosis patient, with an AAV vector expressing the selective marker gene neo from the AAV p.sub.5 promoter was reported (Flotte et al., 1992, Am. J. Respir. Cell. Mol. Biol. 7:349-356). In this study the AAVneo vector was packaged into AAV particles using the pAAV/Ad packaging plasmid. Up to 70% of the cells in the culture could be transduced to geneticin resistance and the particle-to-transducing ratio was similar to that reported by Samulski et al., (1989). Thus to obtain transduction of 70% of the cells, a multiplicity of up to several hundred vector particles per cell was required. Transduction of human airway epithelial cells in in vitro culture using an AAV transducing vector that expressed the CFTR gene from the AAV ITR promoter showed that the cells could be functionally corrected for the electrophysiological defect in chloride channel function that exists in cells from cystic fibrosis patients (Egan et al., Nature, 1992, 358:581-584; Flotte et al., J. Biol. Chem. 268:3781-3790).
The above-cited studies suggest that AAV vectors may have potential utility as vectors for treatment of human disease by gene therapy. However, the ability to generate sufficient amounts of AAV vectors has been a severe limitation on the development of human gene therapy using AAV vectors. One aspect of this limitation has also resulted in the prior absence of any studies using AAV vectors in in vivo animal models. This is generally a reflection of the difficulty associated with generating sufficient amounts of AAV vector stocks having a high enough titer to be useful in analyzing in vivo delivery and gene expression. One of the limiting factors for AAV gene therapy has been the relative inefficiency of the vector packaging systems that have been used. Because of the lack of cell lines expressing the AAV trans complementing functions, such as rep and cap, packaging of AAV vectors has been achieved in adenovirus-infected cells by co-transfection of a packaging plasmid and a vector plasmid. The efficiency of this process may be limited by the efficiency of transfection of each of the plasmid constructs, and by the level of expression of Rep proteins from the packaging plasmids described to date. Each of these problems appears to relate to the biological activities of the AAV Rep proteins. In addition, and as noted above, all of the packaging systems described above have the ability to generate wild-type AAV by recombination.
The lack of cell lines stably expressing functional Rep apparently reflects a cytotoxic or cytostatic function of Rep as shown by the inhibition by Rep of neo-resistant colony formation (Labow et al., 1987; Trempe et al., 1991). This also appears to relate to the tendency of Rep to reverse the immortalized phenotype in cultured cells, which has made the production of cell lines stably expressing functional Rep extremely difficult. Several attempts to generate cell lines expressing Rep have been made. Mendelson et al., (1988, Virology, 166:154-165) reported obtaining in one cell line some low level expression of AAV Rep52 protein but no Rep78 or Rep68 protein after stable transfection of HeLa or 293 cells with plasmids containing an AAV rep gene. Because of the absence of Rep78 and Rep68 proteins, vector could not be produced in the cell line. Another cell line made a barely detectable amount of Rep78 which was nonfunctional.
Vincent et al. (1990, Vaccines 90, Cold Spring Harbor Laboratory Press, pp. 353-359) attempted to generate cell lines containing the AAV rep and cap genes expressed from the normal AAV promoters, but these attempts were not successful either because the vectors were contaminated with a 100-fold excess of wild-type AAV particles or because the vectors were produced at only very low titers of less than 4.times.10.sup.3.
In an alternate approach, Lebkowski et al. (U.S. Pat. No. 5,173,414, issued Dec. 22, 1992) constructed cell lines containing AAV vectors in an episomal plasmid. These cell lines could then be infected with adenovirus and transfected with the trans complementing AAV functions rep and cap to generate preparations of AAV vector. It is claimed that this allows higher titers of AAV stocks to be produced. However, in the examples shown, the only information relative to titer that is shown is that one human cell line, K562, could be transduced at efficiencies of only 1% or less, which does not indicate high titer production of any AAV vector. In this system the vector is carried as an episomal (unintegrated construct), and it is stated that integrated copies of the vector are not preferred.
The approach to packaging of AAV vectors described by Lebkowski et al., 1992, has several undesirable aspects. First, maintaining the vector as an unintegrated, high copy number episomal plasmid in a cell line is not desirable because the copy number per cell cannot be rigorously controlled and episomal DNA is much more likely to undergo rearrangement leading to production of defective vectors. Secondly, in this system, the vector must still be packaged by infecting the cell line with adenovirus and introducing a plasmid containing the AAV rep and cap genes. The plasmid used by Lebkowski et al., 1992, again was pBa1 which, as noted above, has overlapping homology with the vector ITR sequences and will result in generation of wild-type AAV. Third, in the pBa1 packaging plasmid used by Lebkowski et al., 1988, 1992, the rep gene is expressed off its homologous p.sub.5 promoter and is thus negatively autoregulated and therefore rep expression is likely to be limited.
The problem of suboptimal levels of rep expression after plasmid transfection may relate to another biological activity of these proteins. There is evidence (Tratschin et al., 1986, Mol. Cell. Biol. 6:2884-2894) that AAV-Rep proteins down-regulate their own expression from the AAV-p.sub.5 promoter which has been used in all of the previously described packaging constructs such as pAAV/Ad (Samulski et al., 1989) or pBa1 (Lebkowski et al., 1988, 1992).